AI and programming classes in Blackburn

The learner builds a small Interval class in Python holding a low and a high value. Adding two intervals adds the lows and the highs; multiplying positive intervals multiplies them. The pamphlet's 80 spindles per jenny is replaced by the range the book itself gives, from 16 to 120. Multiplied by the 20,070 hand-jennies, the total becomes anywhere from 321,120 to 2,408,400 spindles, a range more than seven times wide.

Mule spindles are 550 times 90, or 49,500. The mule's share of all spindles is 3.0 per cent at the pamphlet's figure, but anything from 2.0 to 13.4 per cent once the jenny size is left open. The tempting shortcut is to take the midpoint, 68 spindles, and report 3.5 per cent. That is a real number with no right to its decimal place, and the program refuses to print it without the range beside it.

Then comes the subtle part, the dependency problem. Suppose the 550 mules were only good to ten per cent. The mule count appears twice in the share, above and below the line. Treat the two appearances as independent and Python reports 2.68 to 3.30 per cent; rewrite the formula so the count appears once and the true range is 2.70 to 3.28. Here the gap is small, but in longer calculations it can swamp the answer. A test checks that x minus x returns exactly zero only when the code knows both are the same x.

Asking how sure an answer is makes a strong case for Blackburn teenagers to keep learning to code in 2026. The longer argument is in why teenagers should still learn to code in 2026.